Rift, Vive, and Virtual Reality

[pjbliverpool]

http://uploadvr.com/valve-cost-vr-graphics/

Very interesting to see that a GeForce 680 will be able to run the vive . The improvements will most likely cross the other side of the pond to oculus also.

A lot of people have 680 level hardware , more to the point in a few months 680 level hardware should cost peanuts. So you could greatly reduce the cost of a vr ready pc from $1200 down to $800 or less

I bloody knew it!! I've been saying since the start that VR will scale down to slower hardware on PC and the 970 won't be the "minimum".

Looks like the PC VR ready installed base is about to get a whole lot bigger, and the cost of entry a whole lot smaller.

It sounds like they are also considering a re-projection solution similar to the PSVR too, but upscaling from 45->90fps. In combination with the resolution scaling they also described it should allow GPU's as slow as the GTX 660 to deliver a "passable" VR experience.

There's a huge yellow band representing performance between red and green on the Steam VR test that says "capable". So people running this test can already see what their hardware achieves.

But the test seems to fail on one major point: most people who run the test and get "capable" seem to think they must upgrade. The text associated with "capable" results is quite unequivocal: "your VR is rubbish, upgrade your graphics". So I'm afraid to say, while "680" performance sounds "inclusive", what's actually going to happen is quite the opposite.

The test doesn't use this resolution scaling at all. It quite clearly says that that my 670 OC (at 680 performance) is totally inadequate for VR, whereas with the resolution scaling detailed above it would be rendering at the native 90fps.

 

[pjbliverpool]

Surely 45 Hz isn't good enough for VR. Are they 'upscaling' like Sony to 90 Hz, or just letting people have a juddery, nauseating experience after spending hundreds of bucks on virtual reality?

I'd say that upscaling is a near certainty. The tech already exists on the PC side to do it (async timewarp) and we've already seen it implemented in the market by Sony. It'd be madness not to do the same on PC.

EDIT: in fact, with async timewarp turned on (and it may be by default, I'm unsure on that), I thin "re-projecting" to 90hz from 45fps would be automatic since that's just how it works, i.e. the whole idea is to fill in "missed" frames using projected head position to ensure 90 unique frames always get to the headset. They'd just need to ensure that GPU's which don't hit the requirement to render at 90fps 99% of the time are automatically capped to 45fps for consistency.

 

[Jawed]

The test doesn't use this resolution scaling at all.

Where is that documented?

It quite clearly says that that my 670 OC (at 680 performance) is totally inadequate for VR, whereas with the resolution scaling detailed above it would be rendering at the native 90fps.

Show us the result from your Steam VR test, including the statistics under the "Show Details" option.

 

[pjbliverpool]

Where is that documented?

It doesn't need to be, the article posted above clearly states that using this scaling method the 680 would run at full frame rate and thus be suitable for VR. But in the test, it's listed as totally unsuitable. Also, the scaling (and possible re-projection) options appear to be works in progress rather than something that is already implemented and working - they say they have "plans in the coming weeks" to update their source code.

Show us the result from your Steam VR test, including the statistics under the "Show Details" option.

I deleted it after it's first use since it's quite large, but I scored marginally better than Ruroini here:

https://forum.beyond3d.com/posts/1896163/

i.e. very near the bottom end of the red part of the spectrum.

 

[Jawed]

It doesn't need to be, the article posted above clearly states that using this scaling method the 680 would run at full frame rate and thus be suitable for VR.

My HD 7970 1GHz scores 1.4 (though it has scored 1.7) and reports 0 to 2 frames below 90fps over about 8500-8900 frames tested. So, I believe the test is using adaptive scaling, because it uses 90fps as a hard threshold for success and then assesses how much adaptive quality loss is applied to achieve 90fps (or better).

This is why Fury X, with merely twice the theoretical capability (bandwidth, compute, texturing, ROPs etc.) of HD 7970, scores substantially more than twice 1.4 (or twice 1.7).

 

[hughJ]

I bloody knew it!! I've been saying since the start that VR will scale down to slower hardware on PC and the 970 won't be the "minimum".

Looks like the PC VR ready installed base is about to get a whole lot bigger, and the cost of entry a whole lot smaller.

For content built on Unity using SteamVR anyways.

Both Oculus and Valve have been approaching their respective VR sdks differently, so I wouldn't hold my breath that Oculus is going to immediately follow suit with everything we see come out of Valve. Valve seem much more willing to have this generation of VR be a gradual R&D rollout that continues to take form while products are in consumer hands, while Oculus seem to be focused on delivering a finished and polished user experience from beginning to end of the product's life cycle. Having the min spec of a subset of VR titles be lower isn't going to change the playing field very much when Oculus's business plan is to provide a customer guaranty that everything on their store will run without issue on X hardware.

What I think might have a bigger impact in the short term (this year) is the fact that certain content built on Unity and running on SteamVR may end up having a substantial visual quality edge for the folks with higher-end systems. This would be a huge competitive advantage between the VR content stores when it comes to influencing the purchasing habits of people that have just shelled out thousands of dollars for VR hardware and high-end GPUs. If the same game can be bought on Oculus's Store and Steam, but looks far better on Steam, then it's not hard to guess where people will gravitate to.

In addition to that you've got the question of what Epic is going to do (if anything) in response. Epic's deferred shading pipeline seems to be unsuited for VR right now (MSAA support being a big one), so much so that Oculus forked UE4 and built a forward renderer to use with their internal ToyBox demo, but all shipping VR content is otherwise still going to be using Epic's vanilla UE4 renderer.

 

[orangpelupa]

Can they decoupled the game render and the VR render?

It sounds weird but that's what happened in Dolphins VR according to this dude on reddit

https://m.reddit.com/r/oculus/comments/49twhh/sony_admits_oculus_rift_is_technically_better/d0v3cov

So the VR tracking is still smooth, can look here look there, even when the game runs at low fps

 

[hughJ]

Sounds like he's decoupling the CPU/client update from the rendering update by resubmitting previous API calls and redrawing the previous game state with updated tracker info? Like a modified display list or something? I dunno, maybe that works in instances where your GPU draw time is next to nothing but I can't see how that system would work with a regular engine.

 

[hughJ]

As an aside - you really know which of your toys you value the most when your neighboring townhouse goes up in flames and you have minutes to fill a sports bag full of equipment and get out. Oculus dk2 got a 3 second glance but didn't make the cut, haha.

 

[tabs]

Very interesting comments by a developer who uses PSVR, Rift and Vive found here.

 

[Silent_Buddha]

Very interesting comments by a developer who uses PSVR, Rift and Vive found here.

Lots of fascinating comments there from the dev.

Rift has the least amount of screen door effect. PSVR has slightly more screen door effect. Vive has more noticeable screen door. This is down to the quality of the optics used. Significantly higher quality optics on the Rift versus PSVR (low quality but has RGB screen) and Vive (lower quality than Rift coupled with pentile screen). Other visual artifacts noted due to the lower quality optics on the Vive.

Lots of FUD spread about the Rift vs. Vive.

Need to play PSVR in a dark room as there is light bleed due to the design (comfort over functionality).

3D sound will diminish greatly if not using the included headphones/ear buds due to differences in driver location versus the included hardware. Developer is expecting Rift to offer a more consistent 3D audio space among users as it's expected many users will ditch the IEMs (due to discomfort) included with PSVR and Vive versus the earphones on the Rift.

Rift has by far the superior headset when it comes to construction and tech and materials.

Rift touch controllers preferred over Vive due to support for finger tracking, more comfortable holding position for hands, and analog grips.

Need special faceplate to use Rift with glasses. Not needed for PSVR or Vive.

Video feed is always a frame or two behind reality making pseudo AR problematic/gimmicky on the Vive. Great for object avoidance and finding your keyboard, however.

Development team feels wide scale adoption of room scale games will be problematic as people are mostly unwilling or unable to dedicate a 3.5m x 3.5m space for it. They find they need to greatly limit the scale of room scale games in order to accommodate players that can't dedicated a lot of space for it. Developer's wishful thinking is that Apartment sizes will increase to drive adoption of said feature. Likely too much trouble to bother with and most developer's may just ditch any plans they have for room scale.

Any game that receives any amount of funding from Sony will be required to sign a contract making that game exclusive (timed or otherwise) to the PSVR.

Currently the majority of studios are focusing on PSVR and Rift. Vive will get some optimized software as well. Everyone else just gets ports of Rift/PSVR versions unless one doesn't exist then they likely get a port of the Vive version. That could change in the future depending on where the market is.

Consumer version of Rift feels like wearing a hat.

So, my personal take away from reading all that.

• Occulus went for a very high quality headset. Optics, weight reducing materials, mechanical IPD adjustment, etc.
• Vive went for a slightly lower quality headset but slightly better room cameras as well as the outfacing camera on the headset.
• PSVR went with a low quality headset. Makes sense since it's meant for the consumer space which is more price sensitive, and it's good enough for what the PS4 can deliver performance wise.

It likely costs Occulus significantly more to manufacture their headset than either of the other two players, especially Sony.

Nothing I read made me want to get any of them. But I do find any and all technology to be fascinating. Especially when it comes to design and design compromises. For example, Sony choosing to focus on comfort (slightly more comfortable than the Rift) at the expense of light bleed into the VR experience (if not in a dark room).

Regards,
SB

 

[eastmen]

https://support.oculus.com/1671802636420076

Shipping dates should be coming soon it seems


What size glasses fit in the headset

https://support.oculus.com/help/oculus/1756184747952053

The FAQ is getting updated , release is 7 days away

 

[eastmen]

https://imgur.com/a/Z67OZ#ODx841l vive retail packaging from bloomberg

 

[Deleted member 13524]

3D sound will diminish greatly if not using the included headphones/ear buds due to differences in driver location versus the included hardware.

Well he never used the word greatly and it sounded more like he's more concerned about people using headsets with their drivers 5cm away from the ears than anything else.
3d sound will always depend a lot on the phase between left and right ears. If the sound source is too far away, you'll mess everything up.


Besides, he clearly wrote that IEMs are best for positional audio:

Positional audio is fine-tuned to sound right on the included IEMs, not on over-ears where the driver is much further away.

Of course, if a person uses a crap $5 headset then the immersion will be crap because the sound is crap nonetheless. But I don't think they will be doing any thorough response frequency analysis to the included headphones and adjust the game's sound output to it.
What would they do with future iterations of headsets with different (better) audio solutions? Let the older games sound worse?


The way I see it, in-ear monitors with a flattest possible response frequency should provide the best possible experience.
If you can afford a quality dual or triple-driver IEM and you feel comfortable with them, you should go with one of those.

 

[Silent_Buddha]

Besides, he clearly wrote that IEMs are best for positional audio:


Of course, if a person uses a crap $5 headset then the immersion will be crap because the sound is crap nonetheless. But I don't think they will be doing any thorough response frequency analysis to the included headphones and adjust the game's sound output to it.
What would they do with future iterations of headsets with different (better) audio solutions? Let the older games sound worse?


The way I see it, in-ear monitors with a flattest possible response frequency should provide the best possible experience.
If you can afford a quality dual or triple-driver IEM and you feel comfortable with them, you should go with one of those.

That was only in reference to replacing the IEM with over the ear headphones. The over the ear driver in that case will be much farther away from the IEM driver, thus delivering the same 3D audio. Here is the quote:

A: No mic jack. We try to work with the included IEMs whenever possible to make sure it will sound on your end once you get a Vive exactly like we meant it to sound. Over-Ears like the MMX 300 might be an issue with the Vive though. Positional audio is fine-tuned to sound right on the included IEMs, not on over-ears where the driver is much further away.

Prior to that he already said the Occulus Rift headphones offer the experience exactly like they meant it to sound (just like the included IEMs), because the audio is tuned for the distance and relative position of the on ear Rift headphones when the Rift is used.

He clarifies things somewhat later.

A: The rifts on-ear headphones sound great and offer us as devs the luxury of being able to target a single device that will sound exactly like envisioned across all Rift users. That's technically still true for the Vive but the fact that they're IEMs will stop many people from using them. At least that's what happened in our office. Out of 10+ people here only one can actually stand IEMs [removed to protect his identity]. They sound pretty good. They're just off the shelf HTC IEMs. Slightly tuned for the beats generation with a lot of low end but not bad at all.

Assuming that each purchaser uses the headphones that are included with the device, you will get the exact audio and positional audio that the developers want. If you replace those then that will no longer be the case. However, in their experience people were inclined to replace their IEMs thus would no longer get the intended audio experience. People in their experience were far less likely to replace their Occulus headphones, and when they did they would often go back to using the Occulus heaphones.

Hence why he believes that the Rift will deliver the developer intended audio experience more faithfully than any of the other devices when taking that into account.

NOTE - more expensive aftermarket audiofile headphones will offer more accurate sound reproduction, but will not offer as accurate 3D positional audio as a trade off. That's what he meant with the response about the IEMs farther up the thread that you were referring to in your reply.

Also it's been known for a while that over ear and on ear generally offer superior audio quality to IEMs not only due to the size of the driver but also due to the fact that the sound will then travel through your ears (with all the folds that affect the sound that reaches your ear) rather than bypassing the majority of your ears. As everyone's ears are different, this isn't something that can be accurately modeled through software or through recording except perhaps on a person by person basis. Something that no game could do (model 7 billion different sets of ears? )

Regards,
SB

 

[Shifty Geezer]

Also it's been known for a while that over ear and on ear generally offer superior audio quality to IEMs not only due to the size of the driver but also due to the fact that the sound will then travel through your ears (with all the folds that affect the sound that reaches your ear) rather than bypassing the majority of your ears. As everyone's ears are different, this isn't something that can be accurately modeled through software or through recording except perhaps on a person by person basis. Something that no game could do (model 7 billion different sets of ears? )

You're grossly exaggerating. Variations in ears aren't that diverse that no two people will have comparable, compatible ear geometry. I doubt they'd need to model more than a few million sets of ears. You could select yours from a drop-down checklist, maybe with a photo icon to match up to your own ear.

 

[pjbliverpool]

Very interesting comments by a developer who uses PSVR, Rift and Vive found here.

Tons of great info there, thanks!

 

[Deleted member 13524]

Also it's been known for a while that over ear and on ear generally offer superior audio quality to IEMs not only due to the size of the driver.

Generally? That's a mouthful. Is it superior for the same price? I doubt that. Is the best over-ear headset better than the best in-ear solution? That's probably subjective, and at least as frequency response goes it's not, AFAIK.
Bigger drivers do give the feeling of "trembling" in the ears on the low range. But if this is what you were talking about, are we still talking about audio quality, since it's perceived by the somatosensory system?

but also due to the fact that the sound will then travel through your ears (with all the folds that affect the sound that reaches your ear) rather than bypassing the majority of your ears.

This makes no sense. The folds you mention are used to attenuate the sound waves differently given the position from where they hit. That's what gives the person an extra chunk of information when they tilt their heads, at least for each new sound that the brain doesn't recognize.
When using headphones the driver is always placed in the same place, the sound waves are always hitting from the same position. In the case of 2-channel headphones, the outer ear is only there to complicate things. Might as well just bypass them using IEMs.

Positional audio should in theory be best served through:
Multi-speaker surround (preferably all of them just full-range) > Multi-speaker Headphones > IEMs > Over-ear > this
On top of all that should be a solution using Dolby Atmos or DTS:X, though I don't think there are any game engines supporting those yet.

Oculus went for over-ear because lots of people don't feel comfortable with IEMs. HTC took a bigger risk with their IEMs in the sense that people may not like to use them, but they probably saved some money and at least have a jack for people to use whatever they want.

As everyone's ears are different, this isn't something that can be accurately modeled through software or through recording except perhaps on a person by person basis. Something that no game could do (model 7 billion different sets of ears?

Or you could just bypass all of it by using IEMs.

 

[hughJ]

Personal experience still tells me that IEMs just dont stack up in practice for delivering that natural sounding expansive audio, and I'm not convinced a bunch of first timer indie devs are going to be engineering an audio experience that can suddenly make an IEM replicate what full sized headphone does dollar for dollar. Even with binaural recordings, which conceivably should be the ideal use case for IEM playback, I don't think they hold a candle to a pair of budget hi-fi headphones.

 

[eastmen]

Wild west outlaws first oculus touch test

Looks pretty cool